p53+/-Rb1-ASCL1low Phenotype Predicts Better Prognosis After Surgical Resection in Large Cell Neuroendocrine Carcinoma: a Molecular Characteristic and Clinical-pathological Investigation of Lung Neuroendocrine Tumors

Background: Based on morphology, necrosis degree, and proliferation status, lung neuroendocrine tumors (NETs) are commonly divided into four subtypes: typical carcinoid (TC), atypical carcinoid (AC), large cell neuroendocrine carcinoma (LCNEC), and small cell lung carcinoma (SCLC). However, cases dicult to classify still exist in diagnosis. Methods: We immunohistochemically investigated the molecular phenotypes in lung NETs and assessed their prognostic value. Results: After morphological re-analysis of 179 NETs, 19 cases were classied as undened, which included 3 cases showing carcinoid morphology with high proliferation (Ki67>20% or mitotic count>10/2mm 2 ), and 16 cases showing intermediate differentiated morphology with Ki67 among 20% to 60%. Furthermore, molecular phenotypes were determined by expression of p53, Rb1, ASCL1, STK11, and gene mutation status of KRAS. Interestingly, p53 wt Rb1 + distinguished an unique subcategory from undened lung NET cases which differ from SCLC or LCNEC in prognosis, indicating WD-NET G3 existed in lung NETs. Additionally, in LCNEC, high ASCL1 expression was only relevant to lymph node metastasis rather than overall survival. However, in p53 +/- Rb1 - LCNEC phenotype, low ASCL1 predicted better outcomes, along with less risk of lymph node metastasis. Conclusion: This study provided evidence for existence of WD-NET G3 in NETs and revealed promising prognosis of p53 +/- Rb1 - ASCL1 low subcategory in LCNEC, which could be benecial to the evaluation of patient status in future clinic practice.


Background
Lung neuroendocrine tumors are considered to originate from the Kulchitsky cells in the bronchial mucosa. In 2015, the World Health Organization (WHO) classi ed lung neuroendocrine tumors into typical carcinoid (TC), atypical carcinoid (AC), small cell lung cancer (SCLC) and large cell neuroendocrine carcinoma (LCNEC) [1]. TC, AC, SCLC and LCNEC can be mostly distinguished by tumor morphology, necrosis degree and mitotic image. Meanwhile Ki67 is usually used as an auxiliary classi cation index [2]. In 2018, the International Agency for Research on Cancer (IARC) and WHO proposed a general classi cation framework for neuroendocrine tumors [3], which jointly de ned neuroendocrine tumors in all organs as neuroendocrine neoplasms. In this framwork, lung well-differentiated neuroendocrine tumor (WD-NET) is divided into G1 and G2, in which G1 equals typical carcinoid and G2 equals atypical carcinoid. LCNEC and SCLC are classi ed into poorly-differentiated neuroendocrine cancer G3 (PD-NEC G3). WD-NET and PD-NEC have signi cantly different clinical characteristics. Generally, WD-NET patients were younger with an average age about 50 years and high 5-year survival rate (TC: 90%−95%; AC: 60% −70%) [4]. The pathogenesis is not closely related to smoking. The patients are usually not sensitive to radiotherapy or chemotherapy (especially platinum), so surgery was preferred. On the other hand, the average age of PD-NEC patients is about 70 years old. PD-NEC is closely related to smoking. Most PD-NEC is sensitive to chemotherapy initially, but the development of acquired chemoresistance is ubiquitous, and the 5-year survival rate is only about 10% [5].
In addition to clinic characteristics, WD-NET and PD-NEC are completely different genetically. According to the whole genome sequencing, carcinoid only has a very low mutation load, and about 11%−13% WD-NET are characterized by MEN1 gene mutation, which is not found in PD-NEC [6]. PD-NEC has high mutation load. In SCLC, the TP53 and RB1 mutation are central events in tumor pathogenesis, with a rate of 90-100% and 92%, respectively [7,8]. According to the different expression of ASCL1, SCLC can be divided into three different molecular subtypes [9,10]: the subgroup with high expression of ASCL1 is called classical subtype, which is the most common; the subgroup with low expression of ASCL1 and high expression of NeuroD1 is called variant subtype; the third is the subtype with negative expression of both genes.
For LCNEC, Rekhtman N et.al. [11] and George J et.al. [12,13] considered LCNEC to be heterogeneous, and could be divided into SCLC-Like subtype, NSCLC-Like subtype and unusual Carcinoid-Like subtype. In SCLC-Like subtype, both TP53 and RB1 mutations exist. In NSCLC-Like subtype, STK11 / KRAS / KEAP1 mutations are common. And Carcinoid-Like subtype has MEN1 mutation with low total mutation load.
George J et.al. [13] further classi ed LCNEC into ASCL1low and ASCL1high subtypes. The ASCL1 high NSCLC-Like cases is called LCNEC-Type I while The SCLC-Like ASCL1 low group is called LCNEC-Type II. Interestingly, they also noticed individual cases diagnosed as LCNEC had carcinoid-like MEN1 mutations, presented carcinoid-like morphology, but with high KI67 LI (> 20%). However, the prognosis of these cases was not studied in their study. This kind of carcinoid tumor with high proliferative activity has also been reported in previous studies, especially in pancreas NENs, namely WD-NET G3. Pancreas WD-NET G3 is different from PD-NEC, the WD-NET G3 patients were not sensitive to chemotherapy such as platinum drugs but had better overall outcomes similar to WD-NET G1/G2. These studies above suggest that further molecular typing of NENs not only result in accurate diagnosis, but also contribute to evaluation of prognosis and therapy selection. In this study, 179 cases of lung neuroendocrine tumors in our hospital from February 2012 to December 2018 were collected. The protein expression of TP53, Rb1, ASCL1, STK11 and the mutation status of KRAS genes were evaluated in TC, AC, SCLC, LCNEC and unclassi ed cases. Based on these markers, we further explored whether WD-NET G3 exists in lung NENs and evaluated the prognosis and clinicopathological characteristics of different lung neuroendocrine tumors after classi cation.

Materials And Methods
Patient cohort 217 cases of pulmonary neuroendocrine tumorfrom February 2012 to December 2018, were collected and sorted retrospectively. According to the following criteria, 179 cases were included in the study.
Case rejection criteria: When patient has both puncture and operation samples, the puncture samples are removed; The patient has only puncture specimen with less tissue, it is not suitable to be sectioned and stained again; The patient was pathologically diagnosed as mixed neuroendocrine tumor of the lung.

Immunohistochemistry and KRAS mutation test
In this study, the staining was conducted on tissue array. 4 um Sections were de-para nized and hydrated, and the endogenous peroxidase was blocked. Antigen retrieval was performed using the Dako Target Retrieval Solution, High PH (Dako Ominis, Agilent Technologies, Santa Clara, CA, USA), in a PTLink set at 98ºC for 25 min. Then slides were incubated with primary antibody for 60 min at room temperature. Immunostaining was achieved by an enzyme-conjugated polymer complex (Dako K8002, Agilent Interpretation standard of the markers above: Under normal conditions, the half-life of p53 protein is very short and it is degraded quickly. The normal peripheral lung tissue was used as a wild-type control group, and immunohistochemical examination showed that the nuclei were scattered and brownish yellow. After mutation, the protein accumulated rapidly and stably located in the nucleus. Immunohistochemistry showed that more than 60% of tumor cells had uniform brownish yellow staining. In addition, TP53 gene can also produce nonsense mutation, and the expression of p53 protein in tumor cells is completely absent compared with the positive control in interstitial tissue. 60%−100% of the tumor cells were positive or negative (indicating mutation), the rest were wild type.
Rb1 encoded protein was located in the nucleus. In the normal cells, the nucleus is brownish yellow. After the gene mutation, the brovascular tissue was used as the positive control, and the expression was negative. Therefore, the expression of nuclear positive staining protein is normal, and the absence of nuclear staining suggests that Rb1 gene mutation. STK11 gene encodes proteins that function in the cytoplasm. In adenocarcinoma, about 33% of STK11 can mutate, resulting in the loss of STK11 protein expression. Therefore, the lung adenocarcinoma was taken as the negative control, the tumor stroma as the positive control. The positive expression of STK11 protein indicated that STK11 was not mutated, and the deletion of STK11 protein indicated that STK11 gene mutation might exist.
As a nuclear transcription factor, ASCL1 protein is expressed in the nucleus, and the staining intensity of the nucleus is evaluated.

Bioinformatical validation
Integrate George Julies DNA/RNA sequencing data set and clinical prognosis data set,we totally obtained 57 case with Complete data information, including 14 cases of sclc-like type. Patients were divided into ASCL1-high/low group according to the mean fpkm.The survival analysis found that the different expression levels of ASCL1 cannot provide a good guide to the prognosis of patients in ovrall LCNEC levels. However, in sclc-like LCNEC, since there are only 14 patients with survival data, it can still be found that patients with high ASCL1 expression seem to have more Poor prognosis. Based on the statistics of our larger sample size, we can signi cantly nd that patients with high ASCL1 expression in sclc-like LCNEC are more likely to have tumor metastasis and a worse prognosis.

Statistical analysis
The software graphpad prism 7.0 was used for statistical analysis. The overall survival (OS) was calculated by the Kaplan-Meier method and compared by the log-rank test. Two groups were compared using Student's t-test, and multiple groups were compared using one-way analysis of variance (ANOVA). Data were presented as the mean ± SEM. P < 0.05 was considered statistically signi cant.
The remaining 19 cases were unde ned-lung NETs. Of them, 3 cases had typical carcinoid morphology ( Fig. 2A): the tumors were arranged in organ like or palisade like arrangement; most of the cells were uniform in morphology; the cytoplasm was medium and light stained; the cell atypia and nucleocytoplasmic ratio were slightly higher than AC; However, the tumor showed high proliferation activity with mitotic image > 10 / 2mm2, or Ki67 ≥ 20% (Fig. 2C&D). In the other 16 cases, the morphology was between PD and WD (Fig. 2D&E): the tumor showed irregular in ltrative growth mode, intratumoral brosis, more tumor necrosis than AC, but the cell nucle-ocytoplasmic ratio was not high, the cell atypia was lower than LCNEC, with trabecula, organ like structure and thin-walled vascular growth mode. Of them, the mitotic image was higher than 10/2mm2 while Ki67 LI was between 20%−60% (Fig. 2F).
Expression of p53 Rb1 STK11 ASCL1 in lung neuroendocrine tumors Representative IHC staining was shown in Fig. 3A-D. The protein expression of p53 and Rb1 was signi cantly different in WD-NET (TC / AC) and PD-NEC (LCNEC / SCLC) ( Table 2). In 29 cases of WD-NET, there was no absence of Rb1 protein, only one case of AC had the absence of tumor p53 expression.

Molecular phenotype of LCNEC and unde ned cases in lung neuroendocrine tumors
According to the expression of p53, Rb1, STK11, ASCL1 and the detection of KRAS mutation, the typical LCNEC and the unde ned cases were re-classi ed (  Immunohistochemical phenotype of p53wtRb1 + with high proliferation avtivity distinguished suspected WD-NET G3 in lung neuroendocrine tumors In the 19 unde ned lung neuroendocrine tumors, there were 8 p53+/-Rb1-cases tted SCLC-Like LCNEC based on morphology. Another 4 p53+/-RB1 + cases had no STK11 loss or KRAS mutation, which were di cult to classify temporarily (Table 3). Importantly, There were 7 Rb1 + p53wt cases, one of which had a KRAS exon 2 c.34G > T (p.G12C) mutation. These p53wtRb1 + cases with high proliferative activity (3 of them have typical carcinoid morphological characteristics) are suspicious as WD-NET G3 subtype. We further analyzed the prognosis of 21 WD-NET, 38 LCNEC, 53 SCLC and 7 suspected WD-NET G3 cases. The survival analysis curve of WD-NET G3 was between WD-NET and PD-NEC. However, the difference did not meet a statistical signi cant level (Fig. 4D).
Low ASCL1 expression in p53+/-Rb1-predicted better outcomes in LCNEC In this study, we analyzed the overall survival rate of LCNEC with different types (Fig. 4D-H). The results showed that there was no signi cant difference in the clinicopathological characteristics and prognosis among SCLC/NSCLC-Like LCNEC and unde ned LCNEC. Among all LCNEC cases, ASCL1 high subgroup was more likely to have lymph node metastasis than ASCL1 low subgroup (Table 4), but there was no signi cant difference in prognosis (Fig. 4G). However, in SCLC-Like LCNEC, SCLC-Like ASCL1 low subtype showed better overall survival compared with SCLC-Like ASCL1 high subtype (Fig. 4G), along with lower risk of lymph node metastasis (Table 5). In addition, we conducted a bioinformatical validation to test the prognosis value of ASCL1 in LCNEC (Fig. 5A) and SCLC-Like LCNEC (Fig. 5B). There only existed one database for transcriptome analysis. The mRNA level of ASCL1 was not in consistence with LCNEC prognosis.

Discussion
Neuroendocrine tumors (NETs) are highly heterogeneous tumors with great differences in the embryo origin, biological behavior and clinicopathological characteristics [14]. In 2010, WHO revised the naming and classi cation of neuroendocrine tumors based on European Neuroendocrine Tumor Society (ENETs) classi cation system [15]. Dependent on the morphological differentiation degree, the NETs are divided into WD-NET and PD-NEC. According to the mitotic and Ki−67 index, the NETs are also divided into low grade (G1), medium grade (G2) and high grade (G3). G3 is also called NEC, including small cells tumors, large cells tumors, and mixed NEC.
According to 2010 WHO classi cation standard, G3 is de ned as NETs with Ki−67 > 20%. But recently, a proportion of neuroendocrine tumors presenting a number of mitoses or a Ki−67 index higher than 20% and a wel-differentiated morphology have been identi ed, calling for a new category, well-differentiated NET grade 3 (WD-NET G3) [16,17,18]. For example, a prospective epidemiological study [19]

Conclusion
Lung NETs, especially LCNEC, can be classi ed by a combined examination of p53, Rb1, STK11, ASCL1 and KRAS. In lung NETs, there may exist a WD-NET G3 subtype, which was previously diagnosed as LCNEC but showed carcinoid-like morphology, high proliferation activity, and is characterized by Rb1 + p53 wt in molecular phenotype. In lung LCNEC, p53 +/− Rb1 − ASCL1 low may represent a novel phenotype with lower risk of lympha node metastasis and longer overall survival time. Above all, the investigation of different molecular types of lung NETs can not only help understand the pathogenesis or classify accurately, but is also advantage of evaluating the prognosis and selecting appropriate therapy. University of Science and Technology. All participants provided informed consent prior to this study.

Availability of data and material
We declare all data are available

Competing interests
The authors declare that they have no con icts of interest.

Funding
This study was supported in part by grants from the National Natural Science Foundation of China ( No. 81770263, 81974016 and 31271040).

Authors' contributions
Yuting Dong performed the experiments and analysed the results. Qian Zhang collected patient follow-up information and records. Guoping Wang, Zitian Huo and Yaqi Duan scored the IHC, Zitian Huo and Yaqi Duan conceived and designed the study. Zitian Huo wrote the paper. All authors have reviewed the manuscript.